ORIGINAL ARTICLE
Full Access

Temporal interactions and moon illumination effect on mammals in a tropical semievergreen forest of Manas National Park, Assam, India

First published: 15 February 2021

Associate Editor: Jennifer Powers

Handling Editor: Antony Lynam

Abstract

Species interactions play a vital role in structuring mammalian communities by stimulating behavioral responses in varied niche dimensions that affect sympatric associations and predator–prey relationships. We determined temporal overlap and effects of the moon cycle on dominant and sub‐dominant mammalian assemblages in Manas National Park, India. A total of 36 species were captured, with 24,865 independent records over 11,294 trap nights. We collected 1,130 photographs of five large‐ and medium‐sized carnivores and 1,541 photographs of 12 small carnivores. Fifty‐one percent of records were detected during diurnal period, followed by 38% in nocturnal phase, and 11% during twilight. Small carnivores such as Prionailurus bengalensis and Viverridae spp. were strictly nocturnal, whereas Martes flavigula and Herpestidae spp. were diurnal. Medium‐sized carnivores were either nocturnal (Neofelis nebulosa) or diurnal (Cuon alpinus), whereas large‐sized carnivores (Panthera tigris, Panthera pardus, and Ursus thibetanus) were cathemeral. A high degree of temporal overlap (>0.75) was found between most sympatric carnivores with distinct activity peaks, while a low overlap (<0.50) was observed between different body‐sized carnivores. Viverrids’ activity was negatively correlated (r = −0.44, p < 0.01) with lunar cycles, perhaps to increase foraging efficiency or as an anti‐predator strategy. Large prey (μ = 133.23°) and small prey (μ = 131.35°) activity were high during brighter nights due to better visual detection in detecting or avoiding predators. Dominant species activity was least affected by the lunar cycle among forest‐dependent mammals, whereas subdominant species activity was either lunarphobic or lunarphilic. The study demonstrates the use of passive camera traps in understanding the behavioral rhythms of tropical mammals.

1 INTRODUCTION

Intraspecific interactions are particularly relevant because of their regulatory role in ecosystems (Schoener, 1974). In many systems, species living in a community are known to segregate in various niche dimensions to coexist (Schoener, 1974). As a means to coexist, animals may also regulate their activity patterns and subsequently increase their fitness (Schoener, 1974). For example, prey species may avoid the times of high predator activity, while predators, in turn, may align their activity patterns according to the prey (Berger & Gese, 2007; Lima & Bednekoff, 1999) or segregate in other niche dimensions, i.e., space and diet (Lucherini et al., 2009). Removal of top predators from a system may also serve to regulate activity and therefore affect species interactions (Caro & O'doherty, 1999). Thus, understanding the way species interact is paramount to detect any modifications in species behavior as well as in designing good practices in conservation (Kamler et al., 2013; Lucherini et al., 2009).

Physical parameters such as climatic and weather patterns are important determinants of animal behavior (Cozzi et al., 2012). For example, the lunar cycle is reported to play a significant role in shaping activity patterns of various species. Several nocturnal animals alter their activity in response to moonlight variation (Clarke, 1983), and, based on the level of such response, one may classify such animals as lunarphobic (Morrison, 1978) or lunarphilic (Gursky, 2003). For some mammals, such as rodents (deer mouse Peromyscus maniculatus, Anderson's gerbil Gerbillus andersoni, greater Egyptian gerbil Gerbillus pyramidum, paca Cuniculus paca) (Clarke, 1983; Kotler et al., 1991; Pratas‐Santiago et al., 2017), armadillos (Dasypus sp.) (Pratas‐Santiago et al., 2017) and bats (Morrison, 1978), brighter nights seem to inhibit activity. Such animals may allocate more activity toward darker nights or darker periods in the night (Daly et al., 1992; Kotler et al., 2002). On the other hand, for other species, such as white‐footed mouse Peromyscus leucopus (Zollner & Lima, 1999), cheetah Acinonyx jubatus (Broekhuis et al., 2014), striped hog‐nosed skunk Conepatus semistriatus (de Matos Dias, 2017), rock cavy Kerodon rupestris (de Matos Dias et al., 2018), some primates (Gursky, 2003), and some nocturnal birds (Wilson & Watts, 2006), brighter nights yield increased activity. This increment of activity is often related to higher predator awareness and increased food uptake success (Gursky, 2003; Wilson & Watts, 2006). Cozzi et al. (2012) stated that subordinate or subdominant species (wild dogs & cheetahs) were found to be more active in full moon nights to avoid interactions with dominant species (lions & hyenas).

Despite much interest in species behavior, the understanding of interactions between tropical species is limited. Tropical systems harbor highly cryptic and secretive species, which presents a challenge for behavioral research (Ridout & Linkie, 2009). However, with the use of passive camera traps, monitoring of such species in tropical forests has aided in generating more useful and reliable forms of data (Cutler & Swann, 1999; Karanth & Nichols, 1998). Numerous studies from South Asian tropical forests have also applied camera traps to demonstrate coexistence, resource partitioning, activity patterns, and temporal interactions in carnivores (Allen et al., 2020; Lynam et al., 2013; Mukherjee et al., 2019; Singh & Macdonald, 2017; Sukmasuang et al., 2020; Sunarto et al., 2015).

We examine how potentially competing species overlap temporally with their sympatrids using a diverse community of mammals in a subtropical forest system. The study region was ideal for examining such aspects as it holds diverse ecological niches and wide variation in body mass of herbivores (2.5–5,300 kg) and carnivores (0.8–200 kg). We expected that larger‐bodied mammals would display longer activity as per van Schaik and Griffiths (1996) and intraguild competition may be reduced if ecologically similar species partition their temporal niche (Di Bitetti et al., 2009; Kronfeld‐Schor & Dayan, 2003), to avoid aggregation risk and competition for resources. Accordingly, we predicted that large sympatric carnivores will show high temporal overlap but would segregate at least during the peaks of activity (Dias et al., 2019; Penido et al., 2017); similarly, the small carnivore guild would separate along a temporal axis, i.e., forming temporally distinct coexistence groups (de Satgé et al., 2017).

Among inter‐classes of body sizes, we hypothesized that temporal overlaps between different body‐sized carnivores would be low as mesopredators would avoid dominant predators, and subsequently, smaller carnivores would avoid predators dominant to them (dominant competitor avoidance hypothesis). Alternatively, however, considering that the activity of predators may be correlated positively with that of their prey (Luttbeg & Sih, 2004), we hypothesized that carnivores’ activity patterns are more synchronized with their preferred prey, especially to the hyper‐carnivorous guild (prey synchrony hypothesis) (Linkie & Ridout, 2011). We investigated the effects of moon cycle as manifestations of interactive effects of dominant species over subdominant counterparts. Increased activity under brighter conditions (visual acuity hypothesis), as well as increased activity under darker conditions (dominant predator avoidance hypothesis), has been interpreted as anti‐predator strategies in mammals (Prugh & Golden, 2014). Accordingly, subdominant predators or prey species activities will be regulated by more dominant species. If this is true, then we expect that these activity rhythms will manifest in the form of variation according to different lunar phases as well with lunar phases either aiding in increased vigilance or avoidance of predation risk.

2 STUDY AREA

The study was carried out in an area of 500 km2 within Manas National Park (MNP) (26°80'–26°50' N, 90°45'–91°15' E), a UNESCO World Heritage Site, in the state of Assam, India. MNP lies on the borders of the Indo‐Gangetic and Indo‐Malayan biogeographical realms on a gentle alluvial slope in the foothills of the Himalayas, where wooded hills give way to grasslands and tropical forest. The climate is subtropical and is characterized by four seasons, viz. (a) pre‐monsoon: March–May, (b) monsoon: June–September, (c) retreating monsoon: October–November, and (d) winter: December–February (Das et al., 2009). The mean temperature ranges from 5 to 37°C. The elevation ranges from 40 m to 200 m a.s.l. (Bhattacharjya et al., 2015). The study site is located in core area of the tiger reserve with no artificial light at night around any locations within the study area. The park is home to a variety of important mammal species, including the tiger (Panthera tigris), pygmy hog (Porcula salvania), hispid hare (Caprolagus hispidus), and Asiatic elephant (Elephas maximus) (Wikramanayake et al., 2002), and also supports 22 of India's most threatened mammal species, as listed in Schedule‐I of the Wildlife (Protection) Act of India (DebRoy, 1991). Together with the Royal Manas National Park in Bhutan, the park forms one of the largest areas for conservation significance in South Asia, representing the full range of habitats from the subtropical plains to the alpine zone (Wang, 2001).

3 METHODS

3.1 Field sampling design

Mammalian carnivores of varying body size, i.e., large (>20 kg), medium (>10–20 kg), and small (1–10 kg) (Ramesh et al., 2015), and trophic positions (Ritchie & Johnson, 2009) along with their potential prey species (Table S1) were photographed using camera traps in an area of 270 km2 within strictly forested habitats of MNP. Camera trap locations were unbaited and selected based on accessibility, terrain features, animal trails, and nallahs (seasonal drainages) with carnivore signs (Marinho et al., 2018; Ramesh et al., 2015). At each location, a single Cuddeback X‐Change™ color model (Cuddeback) with motion sensors and time lag of 1s was set between animal detections. Cameras were fastened to trees at the height of approximately 30–45 cm above the ground for an average of 24 days (range: 16–45). In total, 475 camera trap locations were utilized in a grid‐based approach (grid size: 1 km2) during three sampling periods: April 2017–June 2017 (n = 101), December 2017–May 2018 (n = 154), and November 2018–May 2019 (n = 220) (Figure 1). These months were specifically chosen for greater accessibility and minimal variance in weather conditions. The average temperature and humidity during the entire sampling period were 24°C (±SE 0.29) and 79.4% (±SE 0.54), respectively (http://timeanddate.com).

image
Map of the study area (MNP) shows camera trap locations (n = 475), drainage, and forest cover. Camera traps were deployed in a grid‐based approach (grid size: 1 km2) during three sampling periods: April 2017–June 2017 (n = 101), December 2017–May 2018 (n = 154), and November 2018–May 2019 (n = 220)

3.2 Activity periods and activity analysis

3.2.1 Activity periods

The date and time printed on the photographs were used to determine diel activity periods of each species. We assumed that the number of camera trap records taken at various times was correlated with the daily activity patterns of mammals (Linkie & Ridout, 2011, Batschelet, 1981). To maintain statistical independence and reduce bias caused by repeated detections of the same individual, one record of each species per half hour per camera trap site was considered an independent detection, and subsequent records were removed (O'Brien et al., 2003). Species were classified as diurnal (<10% of observations in the dark), nocturnal (>90% of observations in the dark), mostly diurnal (between 10% and 30% of observations in the dark), mostly nocturnal (between 70% and 90% of observations in the dark), crepuscular (≥50% of observations during the crepuscular phase), or cathemeral (between 30% and 70% photographs in the dark) (Theuerkauf et al., 2003). Photographs that captured an hour before and after sunrise and sunset, respectively, were classified as crepuscular (Linkie & Ridout, 2011). Sunrise and sunset hours were determined using geographical coordinates using Moonphase SH software (version 3.3; Henrik Tingstrom). Camera traps only recorded ground‐level activity.

3.2.2 Activity analysis

The activity patterns for each species were analyzed using Kernel density estimation curves, a nonparametric way to estimate the probability density function of distribution of records which assumes that an animal is equally likely to be captured at any time as long as it is active (Linkie & Ridout, 2011). Overlap coefficients among the daily activity patterns of sympatric carnivores and potential prey were estimated using overlap package (Linkie & Ridout, 2011) in R‐software version 3.1.2 (R Development Core Team, 2011). Overlap coefficient (∆) was defined as the area under the curve that is formed by taking at least two density functions at each time point ranging from 0 (no overlap) to 1 (complete overlap), where 0 implies that the species have no common active period and 1 implies that the activity densities of two species are identical (Schmid & Schmidt, 2006). Schmid and Schmidt (2006) proposed five general nonparametric estimators (Δ1 to Δ5) of the coefficient of overlapping. For circular distributions, the first two (Δ1 & Δ2) are equivalent and the third (Δ3) is unworkable (Ridout & Linkie, 2009). ∆5 is unstable as small incremental changes in the data produce discontinuous changes in the estimate, and it can give estimates greater than one and hence is not useful in any case (Schmid & Schmidt, 2006). The chosen estimator for computing overlap was Δ1 and Δ4, depending upon the sample size. Δ4 estimator for the coefficient of overlap was used if both samples were larger than 50, whereas Δ1 was used for small sample sizes (Linkie & Ridout, 2011). The precision of this estimator was obtained through 95% confidence interval (CI), as percentile intervals from 999 bootstrap samples (Linkie & Ridout, 2011; Meredith & Ridout, 2014).

Because the coefficient of overlap is purely descriptive, we used the Mardia–Watson–Wheeler (MWW) test (Batschelet, 1981) to compare statistically the distribution of detections across the diel cycle for all sampling campaign pairs (Brook et al., 2012; Gerber et al., 2012) in program Oriana (Kovach ,2011). The null hypothesis of a common distribution was rejected if the value of W was larger than the critical value indicated by p < 0.05 (Pewsey et al., 2013). Pairwise comparisons between predator and prey were made using the literature provided in Table S1.

3.2.3 Moon phase

We classified the moon phase of records, according to the percentage of visible moon surface into four categories: 0%–25% [New Moon (New)], >25%–75% [first quarter—Waxing Moon (Wx); last quarter—Waning Moon (Wn)], and 75%–100% [Full Moon (Full)] using Moonphase SH software, version 3.3. The records from each moon phase were selected to assess the effect of moonlight and positioning on the time schedules of large and small carnivores and their potential prey, during the lunar cycle. One‐way ANOVA, Dunn–Sidak, and correlation tests were conducted to measure the degree of association and pairwise comparisons among records of predator–prey as well as dominant–subdominant assemblages in each moon phase.

We examined the effects of lunar cycle on mammalian activity in more detail by computing circular statistics in program Oriana (Kovach, 2011). We assigned each observation night a numerical value, calculated as the days since new moon divided by 29.5 (where 0 represents the new moon, and the number of days in the lunar cycle is 29.5). Then, we multiplied the results obtained by 360° (0°=360°=new moon; 180°=full moon) (Grant et al., 2009). We performed the transformation of lunar cycle to a 360° scale to analyze mammalian activity data through circular statistics using circular–linear correlation (Mardia & Jupp, 2000). We provided the mean vector (μ) and the length of the mean vector (r). The length of the mean vector (r) is a measure of angular dispersion (similar to standard deviation); its value can range from 0 to 1. In terms of our study, a high r‐value denotes that activity of mammals is limited to a particular lunar phase, while a low r‐value indicates that activity is distributed across the lunar cycle. The Rao's spacing test (U) for uniformity around the circular space was used to assess whether activity of mammals was uniform across the lunar cycle (Batschelet, 1981). The Rao's test is more powerful and robust than many other circular goodness‐of‐fit‐tests and is able to analyze bimodal and multimodal distributions, whereas other tests, such as the Rayleigh test and Watson's U2, cannot (Bergin, 1991). The circular histograms we presented in conjunction with these analyses showed the frequencies of mammalian records; therefore, the scales vary depending on the value of the sample size in each case. Lunar cycle analysis was done combined for large carnivores (tiger, leopard, and clouded leopard), large prey, and small prey. For small carnivores, all civets were grouped, and leopard cat was considered as a small felid.

4 RESULTS

A total of 36 species were recorded with 24,865 independent records over the entire sampling period of 11,294 trap nights (Table S2). The independent records for the photo‐captured species varied from clouded leopard (n = 21) to tiger (n = 466) for large–medium carnivores, from Chinese pangolin (n = 2) to small Indian civet (n = 40) for small carnivores, from chital (n = 1) to Asiatic elephant (n = 6,387) for large herbivores, and from hispid hare (n = 8) to Indian hare (n = 92) for small herbivores.

4.1 Activity periods

Activity periods for 36 species indicated no sharp boundaries between categories (diurnal, nocturnal, cathemeral, or crepuscular) (Table S2). Small carnivores were found to be either mainly nocturnal (leopard cat and civets) or diurnal (yellow‐throated marten and mongooses). Medium‐sized carnivores (clouded leopard and dhole) were found to be nocturnal or diurnal, and large‐sized carnivores (tiger, leopard, and Asiatic black bear) were cathemeral. Herbivores were either cathemeral (Bos gaurus, Bubalus arnee, and Rusa unicolor) or diurnal (Elephas maximus, Muntiacus muntjak, and Sus scrofa). Terrestrial birds (red junglefowl, Kalij pheasant, Indian peafowl) were found to be diurnal, whereas the routine activity of Indian hare and Himalayan crestless porcupine suggested nocturnal activity.

4.2 Activity pattern and temporal overlaps

4.2.1 Activity overlaps within trophic level

We detected eight small carnivores and evaluated their diel activity pattern (Figure 2). Small carnivores segregated their temporal niche by forming two temporal groups, i.e., nocturnal (leopard cat and civets) and diurnal (yellow‐throated marten and mongooses). Nocturnal small carnivores exhibited high temporal overlap, with the greatest overlap between small Indian civet and Asian palm civet with an overlap coefficient, Δ4 = 0.92 (95% CI 0.84–0.94); and the MWW test showed significant similarity between their daily distribution of records (W = 1.03, p = 0.6). Similarly, high overlap was also observed between diurnal small carnivores, with the highest coefficient value of Δ1 = 0.85 (95% CI 0.61–0.88) between crab‐eating mongoose and yellow‐throated marten, and not statistically different (W = 1.50, p = 0.47). However, each diurnal–nocturnal species pairing showed significant differences in circadian activities. All small carnivores followed a bimodal activity pattern except for small Indian mongoose that showed a unimodal activity.

image
Overlap of activity patterns between species pairs of small carnivores (within trophic level) in Manas National Park, Assam, India. Individual photograph times are indicated by the short vertical lines above the x‐axis. The overlap coefficient (Δ14) is the area under a minimum of two density estimates, as indicated by the shaded area in each plot. Solid lines represent species in each column; dotted lines represent species in the row. High temporal overlaps (Δ > 0.75) are bold. The W value with an asterisk (*) indicates p < 0.05, and the two sets of circular distributions came from a different distribution. Small carnivores segregated their temporal niche by forming two temporal groups, i.e., nocturnal (leopard cats and civets) and diurnal (yellow‐throated marten and mongooses)

Among large carnivores, tigers and leopards showed the highest daily activity overlap [Δ4 = 0.87 (95% CI 0.81–0.91)]; however, the MWW test showed significant dissimilarity between their diel activity patterns (W = 23.59, p < 0.01) (Figure 3). Both the species were active throughout the day and night, but leopards were more active during daylight, with peaks in the early morning and late afternoon, whereas tigers were least active from 1,000 to 1500 hr. Two activity peaks (from 1900 to 2300 hr and from 0200 to 0400 hr) were observed for clouded leopards, suggesting a bimodal activity pattern of the species. Dholes showed a unimodal pattern of activity with peaks at 0600 hr. A synchronized least‐active pattern was noted at midnight for clouded leopard, dhole, and Asiatic black bear.

image
Overlap of activity patterns between species pairs of large–medium carnivores (within trophic level) in Manas National Park, Assam, India. Individual photograph times are indicated by the short vertical lines above the x‐axis. The overlap coefficient (Δ14) is the area under a minimum of two density estimates, as indicated by the shaded area in each plot. Solid lines represent species in each column; dotted lines represent species in the row. High temporal overlaps (Δ > 0.75) are bold. The W value with an asterisk (*) indicates p < 0.05, and the two sets of circular distributions came from a different distribution. Most species pairs exhibited a high activity overlap, but almost all carnivores segregated at least their activity peaks

4.2.2 Activity overlaps among trophic level

Among inter‐guild interactions, most small carnivores showed moderate temporal overlap (0.54) with their dominant competitors (Figure 4). However, clouded leopard and dhole had moderately high overlap with nocturnal and diurnal small carnivores, respectively, as they were more active during those phases.

image
Overlap of activity patterns between species pairs of large–medium carnivores and small carnivores (among trophic level) in Manas National Park, Assam, India. Individual photograph times are indicated by the short vertical lines above the x‐axis. The overlap coefficient (Δ14) is the area under a minimum of two density estimates, as indicated by the shaded area in each plot. Solid lines represent the activity of large carnivores; dotted lines represent the small carnivores’ activity. High temporal overlaps (Δ > 0.75) are bold. The W value with an asterisk (*) indicates p < 0.05, and the two sets of circular distributions came from a different distribution. Most lesser carnivores showed moderate temporal overlap with their dominant competitors. Clouded leopard and dhole had moderately high overlap with nocturnal and diurnal small carnivores, respectively

4.2.3 Carnivore and its potential prey

Tigers showed the highest temporal overlap with wild buffalo [Δ4 = 0.87 (95% CI 0.82–0.92)], sambar (0.86), and gaur (0.79) (Figure 5). In case of leopards, greatest overlap was found with wild buffalo [Δ4 = 0.81 (95% CI 0.77–0.87)] and gaur (0.81), whereas clouded leopards had maximum overlap with Himalayan crestless porcupine [Δ1 = 0.77 (95% CI 0.58–0.87)] and barking deer (0.49). High temporal overlap was found among diurnal prey species such as barking deer [Δ4 = 0.65 (95% CI 0.61–0.74)] and wild boar (0.63) with dholes, whereas Indian hares were active during night time and showed high overlap with leopard cats. MWW test showed similar activity between tiger and wild buffalo (W = 1.20, p = 0.55), clouded leopard and Himalayan crestless porcupine (W = 5.59, p = 0.06), and leopard cat and Indian hare (W = 4.13, p = 0.13).

image
Pairwise temporal overlap (Δ14) between carnivores [(a) tiger, (b) leopard, (c) clouded leopard, (d) dhole, and (e) leopard cat] and their potential prey in Manas National Park, Assam, India. Individual photograph times are indicated by the short vertical lines above the x‐axis. The overlap coefficient (Δ14) is the area under a minimum of two density estimates, as indicated by the shaded area in each plot. Solid lines represent the activity of carnivores; dotted lines represent the prey species’ activity. High temporal overlaps (Δ > 0.75) are bold. The W value with an asterisk (*) indicates p < 0.05, and the two sets of circular distributions came from a different distribution. Tiger and leopard showed the highest temporal overlap with wild buffalo. Dhole, clouded leopard, and leopard cat showed the highest temporal overlap with barking deer, Himalayan crestless porcupine, and Indian hare, respectively

4.2.4 Moon phase

The activity patterns of mammals were correlated with different moon phases, with the greatest records of large carnivore (n = 939), large prey (n = 21,212), and small prey (n = 92) captured in the full moon, whereas civets (n = 932) and small felid (n = 377) had more photographs during the new moon (Figure 6, Figure S1).

image
The proportion of nocturnal records of (a) large carnivore, (b) civets, (c) leopard cat, (d) large prey, and (e) small prey in different moon phases in Manas National Park, Assam, India. The bars indicate species records in different moon phases. The dashed line separates large carnivores, civets, leopard cats, large prey, and small prey. Civets showed significantly more records during the new moon than in the full moon. Large prey showed greater distribution during the full moon than in the new moon

One‐way ANOVA result indicated significant variance for all the species guild except for large carnivores (Table S3). Dunn–Sidak test showed significantly more records for civets during the new moon (mean differences = 2.92, p < 0.005) than in the full moon. More records of large prey were recorded during the full moon (mean differences = 7.3, p < 0.05) than in the new moon.

The results of partial correlation depicted a negative relation between civets and moon visible surface while controlling for large carnivores (= −0.45, p < 0.05) (Table S4). Bivariate correlation also showed a negative correlation between them (r = −0.44, p < 0.01) (Table S5). Large prey (r = 0.15, p < 0.05) and small prey (r = 0.13, p < 0.05) showed positive correlation with the moon visible surface.

Results of circular statistics were also in line with the above results (Figure 7, Table 1). Civet activity was not uniform during the lunar cycle (U = 333.83, p < 0.01) and was statistically more likely around the new moon (μ = 23.99°, r = 0.45). Large carnivore (μ = 152.86°), large prey (μ = 131.35°), and small prey (μ = 133.23°) were more likely to be active around waxing gibbous and full moon, whereas small felid (μ = 80.55°) records were highest just before the first quarter.

image
Circadian distribution of mammalian activity during the lunar cycle. The circular histograms with grey bars show the distribution of mammalian activity, mean vector (μ), and vector length (r) on days 0–29 following the start of the new moon. The symbols around the outside indicate the eight phases of the moon corresponding to the angles at their position on the circumference (new moon, waxing crescent, first quarter, waxing gibbous, full moon, waning gibbous, third quarter, and waning crescent). The mean vector (μ) and vector length (r) are indicated by black arrow. Civets were more active toward the new moon. Small felid records were highest just before the first quarter. Large carnivores, large prey, and small prey were more active around waxing gibbous and full moon
TABLE 1. Linear–circular association (mean vector, length of mean vector, and Rao's spacing test) between the activity of mammals and lunar cycle
Mean vector Length of mean vector Rao's spacing test
μ r p U
Large Carnivore 152.86° 0.08 0.11 313.87*
Large prey 131.35° 0.20 <0.001 357.20*
Civets 23.99° 0.45 <0.001 333.83*
Small felid 80.55° 0.16 <0.001 294.12*
Small prey 133.23° 0.17 <0.001 214.32*

Note

  • The non‐random distribution (p < .01) was indicated with an asterisk (*).

5 DISCUSSION

We determined temporal ecology of large–small carnivores, including intraguild, predator–prey, and dominant–subdominant interactions, in tropical forests of MNP. Baseline information on mammals with respect to activity patterns, temporal niche partitioning, and the lunar cycle effects in forested habitats of NE India is provided as well. We determined how potentially competing species overlap or avoid each other temporally. The study also provides novel insights into how moonlight influences activity of mammals in the forested habitats of MNP.

5.1 Activity periods

van Schaik and Griffiths (1996) suggest that smaller mammals (<10 kg) tend to be specifically nocturnal or diurnal as an anti‐predation strategy, whereas larger mammals (>10 kg) are more cathemeral because of energy requirements and associated feeding commitments. In this study also, all recorded small‐ and medium‐sized mammals were found to be mainly nocturnal or diurnal and larger‐sized mammals were found to be active during both day and night hours.

5.2 Activity patterns and temporal overlaps

We found differences in tiger and leopard activity peaks, but significant time overlap between them was evident. They avoid competition by hunting with different activities and in different vegetation strata, as well as on different prey species (Seidensticker, 1976). Tigers may use wide and established trails for movement (Kawanishi, 2002), but leopards may avoid them by using smaller trails or different activity patterns. Our study showed considerably higher activity overlaps (>0.50) of tigers with medium‐ to large‐sized prey (Hayward et al., 2012) and leopards with small–medium‐ to large‐sized prey (Hayward et al., 2006). Leopard activity was diurnal (51%) in tandem with medium‐sized prey, i.e., barking deer and wild boar (Hayward et al., 2006). Asiatic black bear was diurnal (Hwang & Garshelis, 2007) or crepuscular (Sharma et al., 2010); however, our study highlighted a cathemeral pattern.

The study showed that clouded leopard activity was predominantly nocturnal (Adul et al., 2015; Azlan & Sharma, 2006; Grassman et al., 2005a; Mukherjee et al., 2019) with bimodal peaks at 1900–2300 hr and 0200–0400 hr. Although overall radiotelemetry studies reported distinct activity peaks in the morning followed by evening crepuscular hours (Austin et al., 2010; Grassman et al., 2005a). We expected low temporal overlap of clouded leopards with tigers and leopards due to the risk of interspecific killing of clouded leopards by both top predators. However, contrary to our expectations, the temporal overlap (>0.70) was relatively high. We believe that clouded leopards may be able to avoid tigers and leopards by exhibiting arboreal behavior when they encounter them (Austin ,2002) or by selecting different prey species (Ngoprasert et al., 2012). The stocky build, large canines, and the large postcanine space make clouded leopards capable of killing relatively large prey (Therrien, 2005). However, our study reported high overlap with small‐ (Himalayan crestless porcupine) to medium‐sized prey (barking deer). Clouded leopard terrestrial activity was found to be less during daytime (n = 2), a behavior predicted for areas that have top predators (van Schaik & Griffiths, 1996). The low capture rate of 21 photographs in 11,294 trap nights in our study does not necessarily reflect low numbers of the felid but rather decreased probability captures along wildlife trails and roads that were frequented by high numbers of tigers and leopards.

Dhole, the only canid species recorded during the study, showed diurnal activity with bimodal peaks, i.e., morning and evening. Their activity was found to be segregated temporarily with dominant competitors (tigers and leopards), similar to most studies in India and Southeast Asia (Grassman et al., 2005b; Jenks et al., 2012; Khoewsree et al., 2020; Singh & Macdonald, 2017; Sukmasuang et al., 2020). However, Nurvianto et al. (2015) and Ghaskadbi et al. (2016) observed a crepuscular activity. The dhole diet includes a wide variety of prey species, ranging from small rodents and hares to gaur (Hayward et al., 2014; Selvan et al., 2013a, b). Their hunting success varies depending on the pack size, prey species, and habitat (Acharya ,2007; Johnsingh, 1983). In tropical semievergreen forests, the species appear to persist in smaller packs and consume medium‐sized prey (Kawanishi & Sunquist, 2008), as smaller packs are more energetically advantageous in the rainforest where large prey species are scarce. Thick vegetation also favors stalk and ambush hunting techniques over cursorial hunting, and competition with tigers and leopards for small‐ to medium‐sized prey is high (Kawanishi & Sunquist, 2004). It was evident that the predator's activity was driven primarily by prey activity; hence, dholes showed high temporal overlap with medium‐sized diurnal prey, such as barking deer and wild boar (Khoewsree et al., 2020; Sukmasuang et al., 2020).

The only small felid to be recorded as strictly nocturnal was leopard cat, which was consistent with other reported studies (Can et al., 2020; Chen et al., 2016; Lynam et al., 2013; Mukherjee et al., 2019). However, few studies contradict the nocturnality of leopard cats (Azlan & Sharma, 2006; Grassman et al., 2005c; Saxena & Rajvanshi, 2014). Civets were active during nocturnal hours, similar to other reported studies (Sukmasuang et al., 2020; Than et al., 2008), although a study of Azlan (2003) suggests crepuscular activity. Mongooses and yellow‐throated marten were documented as diurnal (Sukmasuang et al., 2020; Zhao et al., 2020); however, Prater & Barruel (1971) and Santiapillai et al. (2000) observed cathemeral activity. The emergence of diurnal and nocturnal groups of small carnivores, a finding reflected at both broad and fine temporal scales, is a reflection of such a partitioning strategy. The Himalayan crestless porcupine was nocturnal, consistent with previous studies by Menon (2003) and Ampeng et al. (2018). Indian hare was predominantly active during crepuscular and nocturnal hours (Molur et al., 2005); however, we reported only nocturnal activity. According to Harmsen et al. (2011), the diel activity of many felids is associated with the activity pattern of their prey. The main reason small cats were nocturnal in the study area could be that rodents and hares, their primary prey, were generally nocturnal (Mukherjee et al., 2016; Xiong et al., 2017), although we could not quantify this through camera trapping.

Smaller carnivores may reduce risk of encountering larger carnivores by temporarily avoiding them. Our study showed that small carnivores separated their temporal axis by forming distinct coexistence groups, i.e., diurnal and nocturnal, and exhibited low temporal overlaps with dominant predators. Among competitors, the difference in body size is likely to trigger behavioral avoidance in subordinate species to avoid interspecific or exploitative competition (Lucherini et al., 2009). Fedriani et al. (2000) found that coyotes (Canis latrans), the largest predator, utilized various habitats and food types and practiced interspecific killing of smaller competitors, thus limiting the distribution of grey foxes (Urocyon cinereoargenteus) and bobcats (Felis rufus). Indeed, body size is important in driving coexistence patterns and suggests that similar dynamics may exist among the small carnivores in our study.

Our results support that temporal segregation facilitates the coexistence of carnivore community in a tropical region, suggesting a partial avoidance that may decrease competition as well as the risk of intraguild predation (Carothers and Jaksić, 1984). Although most species pairs in the large carnivore guild exhibited a high overlap, almost all carnivores segregated at least their activity peaks. However, small carnivores exploited different temporal niches. Results also support the dominant competitor avoidance hypothesis. The high temporal overlap among predator–prey might be the result of predators’ effort to synchronize their activities with those of their potential prey species, supporting the prey synchrony hypothesis.

5.3 Do species respond differently to moonlight?

Moonlight has usually been thought to increase predation risk by enhancing the ability of predators to detect prey (Kotler et al., 1991; Nash, 2007), leading to decreased activity or shifts in prey foraging efficiency in brighter nights (Daly et al., 1992; Orrock et al., 2004; Prugh & Golden, 2014). Both increasing activity under brighter conditions, as well as increasing activity under darker conditions, have been interpreted as anti‐predator strategies in mammals. Brighter nights may help prey spot approaching predators, while dark conditions may allow them to remain undetected by their predators.

The results across 17 lunar cycles suggested that large carnivores (r = 0.08) were active non‐differentially across the lunar cycles, and their activity reached its peak every 4–5 days, followed by a gradual decrease. Moonlight strongly influenced the foraging patterns of prey species. Both large‐ and small‐sized prey activities were found to be more during the full moon nights, suggesting that the full moon could increase their visual efficiency in detecting and avoiding predators (Clarke, 1983; Griffin et al., 2005). This anti‐predator behavior is already well recognized for marsupials and rodents (Bitetti et al., 2006; de Matos Dias et al., 2018). Civets mainly use tactile, olfactory, and aural senses while foraging (Mudappa 2013), and their activity is found less during brighter nights (Ampeng et al., 2018; Grassman et al., 2005d). We also found that moonlight negatively influenced civet activity, and their activity decreases with increasing moonlight intensity. Civets were more active during darker nights, possibly to avoid dominant predators and perhaps to increase foraging efficiency.

The moonlight results partially support the dominant predator avoidance hypothesis, while prey species are found to be more active during brighter nights to locate food and predators according to the visual acuity hypothesis (Prugh & Golden, 2014).

6 CAVEATS

Though camera traps effectively recorded temporal patterns, the drawbacks of small data sets for some species still exist. Some species, such as the clouded leopard, small prey, and primate species, were partly or predominantly arboreal, and the activity patterns of such species can be better explained if camera traps are deployed in species‐specific macro‐habitats. Limitations, such as the inability to account for detection probabilities, which are bound to vary with species, camera trap methods, or species’ behavioral responses to traps, should be considered (Harmsen et al., 2010; Ramesh et al., 2012). Our primary focus on mammal activity was targeted explicitly at forested areas of the MNP. We were not able to determine foraging success or risk avoidance. Future large‐scale telemetry methods could be to generate such non‐consumptive associations.

Despite these drawbacks, we found that predators can co‐occur in the absence of pronounced temporal partitioning due to differences in predator–prey body size and prey activity. Temporal partitioning is likely to play an important role in facilitating carnivore coexistence. However, given that interspecific interactions between species and within guilds are multidimensional, further research into spatial and dietary partitioning by these predators will help understand explicit coexistence patterns within the carnivore community.

ACKNOWLEDGMENTS

We thank the Director and Dean, Wildlife Institute of India. We are grateful to (Lt) Doyil Vengayil, Syed Asrafuzzaman, and Harish Kumar, Department of Science and Technology, Government of India, for financial assistance to project on clouded leopards under grant No: EMR/2015/000085 of 01‐04‐2016. We thank the Forest Department, Government of Assam, and Bodoland Territorial Council (BTC) for permitting us to carry out the survey. We are grateful to Mr Hiranya Kumar Sarma (IFS, former Field Director, Manas Tiger Reserve), Mr Amal Chandra Sarma (IFS, Field Director, Manas Tiger Reserve), Mr Abbas Dewan (ACF, Manas Tiger Reserve) for facilitating to work in Manas National Park. We thank the range officers of the MNP, Babul Brahma, Kunja Basumatary, Kameshwar Boro, Pranab Das and the frontline staff for providing logistic support during the field surveys. We are thankful to (Paniram, Dipen, Dipul, Umesh, Bablu, Suraj, Bhadreshwar, and Kangkan) to support field data collection. Tejas, Chiging, Meban, Saurav, Nikunj, and Krishna for their assistance in the field. We appreciate our reviewers and editors for their valuable comments and suggestions for improving our manuscript.

    CONFLICT OF INTETREST

    The corresponding author confirms on behalf of all authors that there have been no involvements that might raise the question of bias in work reported or in the conclusions, implications, or opinions stated.